Tropical climate and vegetation cover during Heinrich event 1: Simulations with coupled climate vegetation models

Abstract

This study focuses on the climate and vegetation responses to abrupt climate change in the Northern Hemisphere during the last glacial period. Two abrupt climate events are explored: the abrupt cooling of the Heinrich event 1 (HE1), followed by the abrupt warming of the Bølling-Allerød interstadial (BA). These two events are simulated by perturbing the freshwater balance of the Atlantic Ocean, with the intention of altering the Atlantic Meridional Overturning Circulation (AMOC) and also of influencing the Intertropical Convergence Zone (ITCZ) and its associated rainbelt. The University of Victoria Earth System-Climate Model (UVic ESCM) is applied in these experiments. The plant-functional types and the temperature from the model output are used for calculating the biome distribution, which is then compared to the available pollen records. In addition, an inter-model comparison for the HE1 is carried out by comparing the UVic ESCM with the Community Climate System Model version 3 (CCSM3). In the UVic ESCM, the HE1 climate is imitated by adding freshwater to the St. Lawrence River where it runs into the North Atlantic Ocean, which causes a slowdown of the AMOC. The weakening of the AMOC is followed by a cooler climate in the North Atlantic Ocean and a warmer climate in the South Atlantic Ocean. This surface temperature see-saw between the Northern and Southern Hemispheres causes a southward shift of the tropical rainbelt. The simulated drier climate north of the Equator during the HE1 event causes an increase of desertification and the retreat of broadleaf forests in West Africa and northern South America. On one hand, the model results for the HE1 event can be shown to be in agreement with the pollen records from tropical Africa and northern South America. On the other hand, the model fails to predict savannah and grassland in western tropical South America. In addition, the model predicts similar biome distributions for the pre-industrial as well as the last glacial climate, except in tropical northern Africa (the Sahel region), western South America, and central North America. These regions are warmer and wetter during the pre-industrial climate compared to the Last Glacial Maximum climate, which means more tropical forest and savannah cover in the pre industrial time period and further extensions of warm temperate forests and grassland during the Last Glacial Maximum. Detailed results of the HE1 experiment are presented in Chapter 3. We intensify the AMOC in our BA event experiment by applying two mechanisms: the addition of freshwater to the Southern Ocean under present-day climate conditions, and the extraction of freshwater from the North Atlantic Ocean against a glacial climate background. Both mechanisms produce a warmer climate in the North Atlantic Ocean and a cooler climate in the South Atlantic Ocean, which leads to a northward shift of the tropical rainbelt. These experiments suggest that grassland, boreal forest, and warm temperate forest could be found in Europe and North America during the BA event. The predictions of grassland on the west coast of North America conforms with the terrestrial records of Walker Lake, showing mostly xerophytic shrubland caused by a dry climate. Another match can be shown for sites in southwestern Europe and the Mediterranean, where steppe forest is mostly recorded. However, the model fails to simulate temperate forest in southeastern North America and in southern Europe. Detailed explanations of these results can be found in Chapter 4. The inter-model comparison suggests that the UVic ESCM using a simplified atmospheric component is still capable of demonstrating the effects of a shift of the ITCZ, although a distinct tropical precipitation pattern is only shown in the CCSM3 with a much more realistic representation of the atmospheric processes. Nevertheless, the changes of the tropical vegetation cover during the HE1 experiment are similar in both models. Grass cover increases in tropical North Africa, and tree cover is reduced in tropical northwest Africa and South America. Correlations and discrepancies between model results and pollen records vary between the two models, except that they both agree over equatorial western Africa and South America. The detailed results are presented in Chapter 5. In summary, the study demonstrates and explains the response of climate and vegetation cover to abrupt climate changes during the last glacial period in the tropics and in the region around the Atlantic Ocean. Furthermore, a direct comparison between model output and pollen records provides new insights in the potential of proxy data-model comparisons.